Want to take part in these discussions? Sign in if you have an account, or apply for one below
Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.
I see, thanks!
That would be useful if such error messages were visible by the user. I’d be particularly interested in getting error messages alerting of duplicate def/prop/theorem-labels…
another example is the entry differential forms on simplices. The redirects to this entry, such as smooth differential forms on simplices, don’t work.
Regarding #101: I completely agree! What I was trying to say was that at the moment the new renderer did not pick up on the fact that the reference was not closed (which is probably fractionally worse than Maruku’s behaviour, which was to give an incomprehensible error). I will try to add some validation when I get the chance, but it is not completely trivial. The only check I can think to make is if the reference exists on the page (and even this won’t work when a page is intended to be included, i.e. the reference is deliberately not present).
I believe I have fixed A first idea of quantum field theory – Fields. The problem was in the mathematics \mathbb{R}^{0 \vert 1}. Maruku refused to accept 0\vert 1 with no space between 0 and vert, but accepted it with a space in between!
Fixed #102 as well. Problem there was a missing $ at the beginning of a tex block. Again, ideally we need to try to detect this, but it’s not completely trivial.
Anyhow, it has been a while since we saw problems with actual renderer, it seems now just that there are issues in the actual source. I will continue rendering the pages that remain to be rendered, and update when that’s done.
Hey, I am so happy! The pages are all good now. And, of course, they load on sensible time scales.
I am sorry, Richard, that this ended up being so much work on your shoulders. But the feel of the Lab is so much more professional now, with the pages loading quickly.
I am wondering if we should somewhere make an official announcement. “Williamson boosts Lab software beyond beta stage” and see if we may combine the celebration with an attempt to attract some new users, thereby. (I am serious here.)
one more entry that must have a source error is formal smooth infinity-groupoid (redirects to it, such as FormalSmooth∞Grpd, do not work)
Re #106: you are very kind, and I very much appreciate the sentiment and encouragement, but I still feel there is a lot to do before I’d feel I’d lived up to such a statement :-). One of the major things I’d really like to tackle is the mathematics rendering.
But we are making progress. When I recover some energy, it will be for example very easy now to add tex-style definition/theorem environments, that people have requested before. I agree that the pages do seem much snappier to load now. We can make them even quicker, but that will be a next step.
I’ll take a look at #107 and other such pages this evening, if that’s OK. Please keep recording such occurrences here! If there is an error when one saves an edit, then I will see the problem in the logs, but I will not see anything if the page does actually save, like in these cases, so it is very good that you pick up on these.
What I usually do, by the way, if I cannot see an obvious problem, is just copy the content of an offending page to the Sandbox, and gradually remove content until the page loads, so that the error can be isolated. Occasionally this can be tricky: the fix in #104 took quite a while to isolate, for instance.
Thanks! Sure, I’ll keep recording issues here, but now there is no rush.
Great! Fixed #107 now after all.
Just wanted to pop by and say thanks to both of you for working so hard on getting things fixed! (Even though all this discussion is happening on a rather inappropriately named thread…)
Thanks Mike! I’ll make a new thread to ’announce’ the renderer once it is pushed to github. Still a few more things I’d like to robustify/add first.
The first of these was to handle the case that links have a line break, which happens surprisingly often. I believe I have now tweaked things so that links with line breaks should be fine. I also tweaked the revision and page creation code to format page names to ensure that there is never more than a single whitespace between words, which is necessary for the way I handle the line breaks to work. There were only a handful of nLab pages which violated this previously, and in several of those cases we had a near duplicate with a single whitespace; I have now merged or removed these.
Next up was to handle the one remaining missing piece of syntax that I am aware, namely for rendering pictures, [[some_picture.jig:pic]]. This was preventing a few dozen pages from rendering, amongst them some fundamental ones like monad, adjunction, Kan extension, and double category. Now done (and those pages rendered).
Just wanted to note that almost all pages are now rendered, there are less than 100 that are not, which need manual intervention. There will no doubt still be some pages with syntax errors which affect rendering, but we can fix these as they go. Before I push to github, I wish to robustify the asynchronous stuff that is involved in the new rendering, and make page saving asynchronous too, so that we can forget about timeouts.
Re 113, once or twice in the past we had a problem of confusion created by a page whose name ended with a trailing whitespace, which IMHO ought also to be forbidden.
Indeed, trailing whitespace is removed as well :-).
caught another entry that is not redirecting: lim^1 and Milnor sequences should have lim^1 redirecting to it, but it doesn’t
A tex block had not been closed correctly. Fixed now.
Fixed a bug (could not handle a / in a page name, as at right/left Kan fibration), and added the possibility to use a different web when linking to a file, as is done in the link to a file on Urs’ web at the bottom of picture of mechanics. Also fixed syntax errors on a few pages that were preventing rendering (there are still a very small number of such pages, which I am addressing gradually).